Page 5 of 6
Re: 2025-26 team win projection contest
Posted: Sat Nov 08, 2025 9:01 pm
by nbacouchside
DQuinn1575 wrote: ↑Sat Nov 08, 2025 7:52 pm
nbacouchside wrote: ↑Sat Nov 08, 2025 3:46 am
Mike G wrote: ↑Fri Nov 07, 2025 5:49 pm
This seems to assume that the 5-2 Bulls will henceforth turn into their predicted 33-win selves, and the 2-6 Mavs will in fact play like a 42-win team the rest of the way; with minor "luck" adjustments.
The season is already about halfway to that 15-game mark where preseason expectations have overall zero bearing on the 82-game outcome. Around that time, only to-date performance is correlated to final win%.
For the record, the analysis I did suggests that at 13 games, Vegas expectations are still worth about 50% of end of season win projection. It drops below 25% after 31 games. It doesn't drop below 5% until after game 51.
I must be missing something, if I do a simple Excel correlation, I get about 61-62% for FanDuel and SportsOdds for the whole season last year, while our entries were in the 64-66% range.
I think you're not understanding the method I'm using (I probably didn't explain it well enough).
What I'm doing is I ran a version of opponent adjusted MOV with HCA included at intervals of every 5 games and then the end of season (82). I also ran the per game difference between the team and their opponents points per game on 3P shots and FTs from what they would have been if they made league average rates. At each of the 5 game intervals, I ran a regression on the values for opponent and HCA adjusted MOV, the 3 point and FT differentials for the team and their opponent, and the pre-season Vegas Wins total to project both end of season Wins and end of season opponent and HCA adjusted MOV. Then I used linear interpolation on the co-efficients for each of those regressions to figure out what the weights should be for every game not included in the 5 game intervals (i.e. after games 6, 7, 8, and 9, etc.).
At 7 games into the season, about 66% of the end of season Win projection comes from Vegas. By game 13, it drops below 50%. By game 31, it's at roughly 25% and then it really trails off from there.
Re: 2025-26 team win projection contest
Posted: Sun Nov 09, 2025 12:15 pm
by Mike G
Improvements averaged .17 to season low errors
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
EExp 5.88 7.24 .47 dtka 6.36 7.93 .38
DRKO 5.91 7.52 .43 vegas 6.38 7.65 .52
Walr 6.13 7.47 .43 Crow 6.86 8.72 .37
bpmW 6.15 7.50 .44 25pr 6.88 8.40 .33
Mgoo 6.15 7.81 .41 WShr 6.91 8.26 .35
perW 6.17 8.42 .33 KPel 6.96 8.42 .40
avgA 6.18 7.67 .42 medi 6.99 8.40 .45
eWin 6.27 8.25 .34 25re 7.31 9.10 .30
knar 6.29 7.73 .37 DQin 8.68 10.44 .25
medi jumped .40, Crow unmoved.
rogue exponential leaders:
Code: Select all
e = .10 e = .30 e = .66 e = 1.86 e = 16
3.62 eWin 4.42 perW 5.27 DRKO 7.07 EExp 12.2 Walr
3.89 perW 4.51 eWin 5.33 perW 7.32 DRKO 12.9 EExp
4.04 DRKO 4.51 DRKO 5.33 EExp 7.32 Walr 13.1 vegas
4.24 Walr 4.71 EExp 5.50 eWin 7.34 bpmW 13.1 WShr
4.24 dtka 4.74 Walr 5.51 Walr 7.49 vegas 13.8 bpmW
Re: 2025-26 team win projection contest
Posted: Sun Nov 09, 2025 7:02 pm
by Crow
BRef now projecting 6 under 30 wins but only 1 under 20. At the moment, I am leaning toward 3 under 20.
Thunder projected at 54 wins. Uh huh...
BRef projecting 10 teams with a 3+% chance at title. 19 with a non-zero chance. Thunder at under 16%. I'll think about that, though it will change.
Heat with a 10% chance? I don't think so. But more to study and watch. 2W-3L against. 500+ Win teams. 9 of next 10 against .500+ teams.
All trivial projected chances, but Hawks with more chance at title than Warriors and Magic combined? Formula, not enough reason. 22nd rated offense. 1W - 5L against .500 and better teams.
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 1:20 am
by Mike G
Thunder projected at 54 wins. Uh huh...
OKC is 9-1 with a single win against a good team (Hou in 2OT on opening night), plus Atl which is 5-4 outside of their OKC game.
In SRS they are a point behind Hou and 2.6 less than Den. This is all without Jalen Williams, and they've adjusted to win more dominantly of late, outside their loss to otherwise 4-4 Por.
Their schedule strength has been -4.2 to date, almost unthinkably weak for 10 games. Next lightest is Cha at -2.95, one of only 4 teams even half as weak.
They are still a very deep team and can weather the occasional and inevitable loss of a key player or 2 or even 3 for a time. So it's likely they will win more than 54, perhaps 60% or so; but there's unlimited ways they could do worse.
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 3:04 am
by Crow
By the team results in schedule so far, 54 wins is projected. But that is unlikely to be close to the end result. In my view. I don't know if BRef's formula looks at player strength or last season but both improve the case for a higher number.
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 2:23 pm
by Mike G
https://www.basketball-reference.com/fr ... _prob.html
This page gives the data used in their projections: Current record and SRS, SOS to date, remaining SOS; and some regression to 41-41. They don't imagine the Celtics still have Tatum, Kornet, Porzingis etc.
Seismic shifts overnight. The avg error rose by .24, but it ranged from +.49 (PER) to -.11 (vegas).
EExp increased their lead over avgA from .30 to .47
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
EExp 5.99 7.51 .46 perW 6.66 8.88 .31
DRKO 6.09 7.75 .43 eWin 6.73 8.61 .33
vegas 6.21 7.77 .52 Crow 6.90 8.87 .37
Walr 6.30 7.68 .44 medi 7.05 8.50 .45
Mgoo 6.44 8.16 .40 25pr 7.15 8.69 .33
avgA 6.52 7.95 .42 WShr 7.19 8.59 .34
bpmW 6.57 7.79 .44 KPel 7.26 8.70 .39
knar 6.62 7.99 .38 25re 7.55 9.38 .30
dtka 6.66 8.11 .39 DQin 8.90 10.69 .25
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 10:19 pm
by dtkavana
https://craftednba.com/season-projections
Throwing this out as an option to use instead of BBREF's projections if anyone would be interested in that.
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 10:54 pm
by Mike G
Using numbers at that link:
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
vegas 4.03 4.97 .83 EExp 5.18 5.96 .78
medi 4.84 6.03 .75 dtka 5.23 5.94 .78
bpmW 4.86 5.98 .77 25pr 5.44 6.90 .68
avgA 4.87 5.92 .78 25re 5.45 6.98 .66
DRKO 4.89 5.76 .80 KPel 5.51 7.10 .66
Walr 4.90 5.88 .79 knar 5.66 6.66 .73
Crow 4.93 6.14 .74 DQin 5.94 7.66 .62
Mgoo 5.07 6.55 .71 eWin 6.05 7.54 .61
WShr 5.16 6.69 .70 perW 6.64 8.15 .54
Errors relative to these projections or predictions as of this day, Nov. 10, 2025:
Code: Select all
west W east W
OKC 66 Cle 56
Hou 59 NYK 54
Den 54 Det 48
Min 51 Mil 46
GSW 46 Bos 45
LAC 46 Mia 42
LAL 45 Orl 42
SAS 44 Atl 41
Por 41 Chi 40
Phx 34 Tor 39
Mem 33 Phl 37
Dal 31 Ind 31
Sac 27 Cha 25
NOP 23 Brk 19
Uta 22 Was 17
Re: 2025-26 team win projection contest
Posted: Mon Nov 10, 2025 11:39 pm
by Crow
Seems reasonable and looks good to me.
Re: 2025-26 team win projection contest
Posted: Tue Nov 11, 2025 8:24 pm
by DQuinn1575
lol makes me look a lot better. Not good, just not horrendous
Re: 2025-26 team win projection contest
Posted: Tue Nov 11, 2025 9:35 pm
by Mike G
And back to the actual season in progress:
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
EExp 5.97 7.52 .47 dtka 6.68 8.11 .40
DRKO 6.11 7.78 .43 eWin 6.74 8.55 .34
vegas 6.19 7.73 .53 Crow 6.93 8.96 .36
Walr 6.40 7.75 .44 medi 7.09 8.62 .44
Mgoo 6.52 8.13 .41 WShr 7.23 8.59 .35
avgA 6.53 7.97 .42 25pr 7.25 8.71 .33
bpmW 6.63 7.82 .44 KPel 7.32 8.71 .39
perW 6.65 8.85 .32 25re 7.55 9.38 .30
knar 6.67 8.00 .38 DQin 8.86 10.68 .25
Two days ago, vegas was 11th in error but with the best r-squared.
UPDATE Nov. 12 -- Virtually no change in the standings, but everyone got worse by .20 to .33 -- the overachieving teams keep overing and the unders keep undering.
Teams have played 9 to 12 games now.
Code: Select all
over tm avgA proj over tm avgA proj
14.1 Mia 37.3 51.4 -17.8 Dal 45.6 27.8
11.6 Chi 35.4 47.0 -17.4 Ind 39.3 21.9
9.7 Por 36.1 45.8 -15.9 LAC 49.3 33.4
8.7 Phl 37.2 45.9 -9.6 Mem 42.8 33.2
8.5 Tor 36.8 45.3 -7.8 NOP 32.8 25.0
7.9 Uta 24.0 31.9 -7.4 GSW 48.3 40.9
7.9 SAS 42.5 50.4 -7.3 Orl 48.9 41.6
5.7 Phx 33.9 39.6 -6.3 OKC 62.7 56.4
5.7 Det 42.8 48.5 -4.5 Cle 53.0 48.5
4.9 Den 50.6 55.5 -4.5 Sac 38.2 33.7
4.1 NYK 48.3 52.4 -3.2 Brk 25.7 22.5
3.8 LAL 44.0 47.8 -1.1 Atl 43.4 42.3
3.5 Cha 30.2 33.7
3.3 Hou 49.7 53.0
2.5 Mil 41.2 43.7
0.3 Min 46.7 47.0
0.2 Bos 43.1 43.3
0.2 Was 20.5 20.7
Teams on the left were 5-1 (Bos) last night; on the right (OKC) 1-5
UPDATE Nov. 13 -- DARKO goes from .18 down to .21 up, overnight.
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
DRKO 6.24 7.79 .47 medi 7.05 8.79 .44
EExp 6.45 7.88 .46 perW 7.09 9.43 .29
vegas 6.62 8.03 .51 eWin 7.19 8.97 .33
Walr 6.76 8.06 .44 Crow 7.20 9.19 .37
avgA 6.76 8.37 .41 WShr 7.43 9.13 .32
knar 6.78 8.17 .41 KPel 7.59 9.43 .34
bpmW 6.93 8.22 .43 25pr 7.76 9.36 .29
Mgoo 6.94 8.63 .39 25re 8.25 10.01 .27
dtka 6.94 8.49 .39 DQin 8.86 11.06 .24
UPDATE Nov. 14 -- no change in standings. Overall worst this month.
Code: Select all
W Atl W Bos W Brk W Cha W Chi W Cle W Det W Ind
46 b-r 50 DQ 33 WS 36 Wa 45 b-r 58 md 51 b-r 47 DQ
46 DQ 48 WS 31 kn 35 b-r 41 Cr 57 DO 46 Cr 46 KP
46 dt 47 b-r 28 bW 34 kn 38 WS 56 dt 46 md 43 WS
45 md 46 DO 28 DO 32 eW 38 KP 55 DQ 45 eW 41 Cr
44 kn 46 Cr 27 pW 31 DO 36 kn 54 WS 44 DO 41 pW
44 KP 46 KP 27 eW 31 pW 36 md 53 Wa 44 WS 40 md
44 pW 43 kn 26 DQ 31 bW 35 DQ 53 bW 43 bW 39 Wa
44 bW 42 dt 26 Wa 31 Cr 35 bW 52 Cr 43 Wa 36 bW
43 eW 42 bW 25 dt 30 dt 35 DO 51 kn 43 KP 36 DO
43 DO 42 md 23 b-r 29 KP 34 dt 50 KP 42 pW 36 dt
42 WS 41 Wa 22 Cr 27 WS 34 pW 50 pW 41 kn 36 eW
42 Wa 39 pW 19 md 26 md 33 Wa 50 eW 41 dt 33 kn
39 Cr 37 eW 18 KP 23 DQ 32 eW 49 b-r 35 DQ 19 b-r
^ Atl ^ Bos ^ Brk ^ Cha ^ Chi ^ Cle ^ Det ^ Ind
W Mia W Mil W NYK W Orl W Phl W Tor W Was
49 b-r 45 KP 52 Wa 55 eW 48 b-r 48 b-r 25 DO
41 eW 44 Cr 51 DO 52 Cr 47 md 41 eW 24 Wa
41 WS 43 Wa 50 pW 51 kn 43 pW 41 KP 23 dt
40 KP 42 eW 50 b-r 50 DO 38 KP 40 DO 23 kn
40 pW 42 dt 49 md 50 md 38 eW 39 kn 22 eW
40 bW 42 DO 49 dt 50 KP 38 Cr 38 pW 21 WS
38 Wa 42 md 48 kn 49 Wa 37 dt 38 bW 21 pW
38 kn 41 b-r 48 Cr 48 pW 37 DO 38 Wa 21 Cr
36 DQ 40 DQ 48 WS 48 bW 35 bW 37 dt 21 b-r
36 dt 39 kn 47 KP 46 dt 34 Wa 35 Cr 19 bW
34 DO 38 pW 47 eW 46 b-r 34 kn 32 WS 18 md
33 md 38 bW 47 bW 43 WS 33 DQ 32 md 17 DQ
29 Cr 38 WS 43 DQ 42 DQ 32 WS 28 DQ 14 KP
^ Mia ^ Mil ^ NYK ^ Orl ^ Phl ^ Tor ^ Was
W Dal W Den W GSW W Hou W LAC W LAL W Mem W Min
54 pW 56 b-r 56 KP 58 DQ 53 dt 47 pW 48 DQ 52 md
51 eW 55 pW 51 DQ 54 b-r 51 Cr 47 dt 47 WS 49 WS
50 DQ 54 md 51 Cr 53 Cr 51 kn 47 eW 46 md 48 DQ
46 Cr 52 KP 50 md 52 DO 50 DQ 46 bW 45 KP 48 Cr
45 bW 52 DO 50 dt 51 md 50 eW 46 KP 43 Wa 48 dt
44 DO 52 DQ 49 bW 50 kn 50 KP 45 b-r 43 dt 47 bW
44 KP 52 Cr 48 kn 50 bW 50 DO 44 WS 43 bW 47 kn
44 WS 51 bW 47 WS 50 Wa 49 pW 44 Wa 42 eW 46 eW
43 kn 50 WS 47 Wa 49 dt 49 bW 43 md 41 pW 46 b-r
43 md 50 Wa 46 eW 47 WS 47 WS 42 Cr 41 kn 46 Wa
42 dt 48 dt 44 DO 47 KP 46 md 41 DQ 39 Cr 45 DO
39 Wa 47 eW 42 pW 46 eW 46 Wa 41 kn 38 DO 44 KP
27 b-r 45 kn 41 b-r 46 pW 34 b-r 39 DO 30 b-r 41 pW
^ Dal ^ Den ^ GSW ^ Hou ^ LAC ^ LAL ^ Mem ^ Min
W NOP W OKC W Phx W Por W Sac W SAS W Uta
36 kn 69 DQ 42 b-r 45 b-r 44 KP 50 pW 31 b-r
35 pW 67 Wa 39 DQ 43 md 42 pW 49 b-r 28 Wa
34 Wa 65 bW 37 WS 39 DQ 40 eW 45 eW 28 DO
34 Cr 65 WS 37 kn 38 WS 40 WS 45 DO 26 kn
33 bW 64 Cr 36 dt 38 bW 39 bW 43 kn 26 dt
33 DQ 64 md 35 KP 37 Wa 38 kn 42 Cr 25 eW
33 dt 63 kn 35 bW 36 DO 37 dt 42 Wa 25 WS
33 KP 61 dt 33 eW 36 kn 36 Cr 42 bW 24 bW
33 md 60 eW 33 DO 36 Cr 36 Wa 40 KP 24 md
32 eW 60 DO 32 pW 36 dt 35 DQ 40 DQ 23 pW
32 DO 59 KP 32 Cr 36 KP 35 DO 40 dt 21 DQ
29 WS 58 b-r 32 Wa 32 eW 34 md 40 md 20 KP
25 b-r 58 pW 27 md 30 pW 30 b-r 38 WS 19 Cr
^ NOP ^ OKC ^ Phx ^ Por ^ Sac ^ SAS ^ Uta
All but 7 teams are outside the limits of guesses, or within a single outlier.
Re: 2025-26 team win projection contest
Posted: Sat Nov 15, 2025 12:53 am
by Crow
I was going to wait til Monday, but got tired of waiting.
This is a blend of 7 pre-season projections doing as least pretty well so far & well in past and the 3 updated projections (BRef, Couchside and CraftedPM) from last few days to offer another projection of final results. The updated projections get a modestly higher total weight than the pre-season group.
I might still update later.
45.5 Atlanta Hawks
44.2 Boston Celtics
21.8 Brooklyn Nets
29.7 Charlotte Hornets
38.5 Chicago Bulls
54 Cleveland Cavaliers
35.9 Dallas Mavericks
53.8 Denver Nuggets
46.5 Detroit Pistons
46.4 Golden State Warriors
53.8 Houston Rockets
30.3 Indiana Pacers
43.5 Los Angeles Clippers
45.8 Los Angeles Lakers
36.7 Memphis Grizzlies
41.4 Miami Heat
42.8 Milwaukee Bucks
48.4 Minnesota Timberwolves
28 New Orleans Pelicans
51.1 New York Knicks
62.3 Oklahoma City Thunder
47.7 Orlando Magic
41.3 Philadelphia 76ers
35.8 Phoenix Suns
40.1 Portland Trail Blazers
32.6 Sacramento Kings
45.8 San Antonio Spurs
41.3 Toronto Raptors
25.4 Utah Jazz
19.6 Washington Wizards
1230
I will also add a set of my own updated projections (based on available information and subjective considerations) by Tuesday.
Re: 2025-26 team win projection contest
Posted: Sat Nov 15, 2025 5:40 pm
by Mike G
There's a sort of understated rivalry between the ESPN Experts' averages and the APBRMetricians' ; this year, EExp has had the better/lower error-apparent every day. It bobs up and down though, and pretty significantly.
Highs and lows so far:
Code: Select all
date EExp < avgA
10.26 .47
10.30 .17
11.01 .35
11.02 .16
11.04 .51
11.07 .10
11.12 .61
11.13 .30
11.15 .51
A big win or loss can change a team's projection by 2 wins or more.
+/- 2W by Atl, Mil, NY, Phl, Was, LAL, or OKC adds or subtracts as much as .14 to these error differences. They've already been closer than that.
Re: 2025-26 team win projection contest
Posted: Sat Nov 15, 2025 7:20 pm
by Crow
Based on my blended projection benchmark, Vegas is closest of the 7 pre-season sets I looked at. Updated Couchside and Crafted projections closest overall but they have big shares in the blend. BRef is furthest set from the blend despite a similar weight advantage.
My draft updated projections are 2nd closest to the blend. My original set, not looking good competitively but much lower errors than when compared to BRef, though that is true for all checked. Everybody below 4.0 "error" compared to my blended benchmark except BRef at 4.3.
All checked benefit from some self-correlatiom. Almost everyone benefits (now) from far less weight on BRef projections.
BRef projections from yesterday vs. most recent versions grabbed from Couchside and Crafted and my blend? I'll see how they do over time and in the end.
Re: 2025-26 team win projection contest
Posted: Mon Nov 17, 2025 12:28 am
by Mike G
Code: Select all
. avg err rmse r^2 . avg err rmse r^2
DRKO 6.60 8.18 .46 Mgoo 7.33 8.99 .38
EExp 6.63 8.25 .46 eWin 7.41 9.31 .33
vegas 6.76 8.26 .51 medi 7.51 9.26 .42
knar 7.15 8.49 .42 Crow 7.61 9.52 .36
avgA 7.16 8.74 .40 WShr 7.73 9.49 .32
bpmW 7.19 8.52 .43 KPel 7.89 9.76 .34
dtka 7.19 8.76 .40 25pr 8.12 9.75 .29
perW 7.24 9.81 .29 DQin 8.99 11.36 .24
Walr 7.25 8.49 .43
2025-regressed has been dropped from the competition. 2025 Pythagorean-regressed has dropped from #3 to next-to-last since Nov. 1